INDEX

Explanations

phrases related to changes or increments

oai_token-act-pair · gpt-3.5-turbo

phrases related to increases and decreases in various metrics or rates

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eren

-0.65

rief

-0.65

orno

-0.63

pb

-0.61

Fun

-0.60

ija

-0.59

love

-0.59

famous

-0.58

view

-0.58

OME

-0.57

POSITIVE LOGITS

 increases

3.28

 decreases

2.83

 Increases

2.17

Increases

2.05

 increase

2.05

 rises

1.95

 reductions

1.87

 Increase

1.83

 boosts

1.83

incre

1.81

Activations Density 0.017%

phrases related to changes or increments

phrases related to increases and decreases in various metrics or rates

No Comments

No Known Activations